Statistical Methods for DNA Methylation Data Abstract The overall objective of this study is to develop novel statistical methods and software to study how DNA methylation profiles associated with cancers. DNA methylation is a molecular modification of DNA that is important for normal organism development. Genes that are rich in CpG dinucleotides are usually not methylated in normal tissues, but are frequently hypermethylated in cancer. This is often associated with gene silencing and is an important mechanism for the inactivation of tumor suppressor genes. Studies have also suggested that methylation profiles differ between cancers arising in different organs and between different cancer histologies from the same organ. For example, different DNA methylation profiles are found in different subtypes of leukemia and lung cancer. With the rapid development in array technologies, high-throughput arrays with DNA mathylation measures on the genome-wide level have become widely available. There is a great need for development of novel statistical models to evaluate complex DNA methylation data generated with high-throughput platforms. The specific objectives of this project are: (1) to develop novel models for the distribution of methylation proportions to select differentially methylated loci between cancer and normal subjects;(2) to propose a new classification method that differentiates tumor subtypes using DNA methylation profiles;(3) to develop computer software packages that implement methods developed in specific aims 1-2. The proposed methods will be applied to an existing data with tumor samples/normal tissue samples and an ongoing methylation study the PI is collaborating. We believe the proposed methods will significantly improve current and future efforts in understanding the significance of DNA methylation profiles in cancers. PUBLIC HEALTH RELEVANCE: Abstract narrative To develop a series of novel and powerful statistical methods to study DNA methylation profiles. The proposed methods will significantly improve current and future efforts in understanding the significance of DNA methylation profiles in cancers.